JP3270356B2 - Utterance document creation device, utterance document creation method, and computer-readable recording medium storing a program for causing a computer to execute the utterance document creation procedure - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prepare information in which two expression types of information, i.e., character information and voice information, are combined consistently by allowing speech data for selected voice to correspond to inputted character- string and storing it as speech text. SOLUTION: A speech data storage part 103 stores speech data by each group by grouping the speech data with the same character-string information on the basis of the character-string information of the speech data. Using the character-string inputted through an inputting part 106, a group having the same character-string information as the character-string is retrieved from the speech data storage part 103. A voice synthesis part 105 successively reads out the speech data in the retrieved group and synthesizes voice using the information read out and the timbre data stored in a timbre data storage part 104. A speech text storage part 110 allows the speech data for selected voice to correspond to the inputted character-string and stores it as speech text.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an utterance document creation apparatus for creating an utterance document by adding, to a character string constituting a document, speaking style data for specifying how each character string is spoken.
The present invention relates to a computer-readable recording medium storing a program for causing a computer to execute an utterance document creation method and an utterance document creation procedure.

[0002]

2. Description of the Related Art As a method of transmitting and storing information, a method using character information is known. Recently, as a method of using this character information, a document creator creates a desired document using a document creating device such as a Japanese word processor or an English word processor, or a personal computer having a word processing function, and transmits the created document to a network. Transfer via
A method of storing a created document in a recording medium such as a magnetic disk or an optical disk has been used. This is due to the development of computer-related technology, which has realized higher functionality and lower cost of the document creation device itself, as well as the promotion of paperless offices, the development of communication networks, and the spread of e-mail. It depends heavily on environmental changes.

As other methods used for transmitting and storing information, a method using audio information and a method using audio information and video information are known. For example, in the method using voice information, voice information is transmitted directly through a telephone or the like, and information is stored and recorded on a tape or the like using a recording device. In the method using audio information and video information, information is transmitted using a communication device having a monitor and a speaker, and audio information and video information are transferred, and information is stored using a recording device such as a video device. Stored on tape, optical disk, etc.

[0004] Among the above-mentioned methods of transmitting and storing information, the method using character information has a smaller amount of data and is easier to edit information than other methods, and furthermore, a computer as digital information. Since it can be used above, it is the most versatile and widely used.

[0005]

However, according to the conventional method using character information, since the created document is information limited to visual linguistic information (that is, character linguistic information), it corresponds to non-linguistic information. Expressions such as emotions could not be added as information. In speech-based linguistic information (ie, speech linguistic information), expressions such as emotions that correspond to non-linguistic information can be expressed by changing the way of speaking, such as accent, loudness (volume), and pitch. Was added as information.

Further, in the prior art, there has not been provided an apparatus and a method for creating information in which information of two expression forms, that is, character information and voice information, is combined in a consistent manner.

[0007] The editing of audio information is basically performed using the auditory sense (that is, by listening to the reproduced audio information by ear).
This is generally performed, and it is necessary to reproduce the data one by one to check the position (location) of the desired audio information, so that there is a problem that the operation is complicated and troublesome.

It is possible to synthesize speech from a text document (ie, character information) by using a text speech synthesis technique, which is one of the conventional speech synthesis techniques. However, there was a problem that proper nouns not included in the dictionary were misread or pronounced with different accents. There are also problems that it is not possible to express emotions, which are non-verbal information, and that it is not possible to synthesize speech accurately according to the way the creator intends to speak.

The present invention has been made in view of the above, and it is possible to create information (an utterance document) in which information of two expression forms, that is, character information and voice information, is combined in a consistent manner. It is an object to provide an apparatus and a method.

Further, the present invention has been made in view of the above, and is based on information (a spoken document) in which character information and speech information (speech data) including a manner of speech intended by a document creator are obtained. ) Is created so that expressions such as emotions corresponding to non-verbal information can be added to a document.

Further, the present invention has been made in view of the above, and it is intended to improve the workability by visually editing audio information through character information, and to improve the way of speaking intended by a document creator. It is an object of the present invention to be able to synthesize speech accurately by using.

[0012]

In order to achieve the above object, an utterance document creating apparatus according to claim 1 comprises a character string information composed of a word, a phrase or a sentence, and a phoneme corresponding to a character of the character string information. , The duration of each phoneme in the phoneme string information, the pitch information specifying the relative pitch of the phoneme string information at an arbitrary point in time, and the volume of each phoneme in the phoneme string information. Based on the character string information of the speaking data, the speaking data having the same character string information and different readings are grouped and stored for each group. Using a character string input means for inputting a character string constituting a word, a phrase or a sentence, and a character string input via the character string input means, Search means for searching a group having character string information from the speech data storage means; voice data storage means for storing a plurality of voice data for adding voice to a synthesized voice; The voice data is sequentially read out, and a voice is generated by using the phoneme string information, the duration time, the pitch information and the velocity information of the read voice data and one of the voice data stored in the voice data storage means. Voice synthesizing means for synthesizing, voice selecting means for selecting a desired voice from the voice synthesized by the voice synthesizing means, and speech data corresponding to the voice selected by using the voice selecting means, Utterance document storage means for storing as an utterance document in correspondence with a character string input via the column input means;
It is provided with.

Further, the speech document creating apparatus according to the second aspect of the present invention is characterized in that character string information composed of words, phrases or sentences, phoneme string information composed of phonemes corresponding to the characters of the character string information, Speaking composed of a duration of a phoneme, pitch information specifying a relative pitch of the phoneme string information at an arbitrary point in time, and velocity information specifying a relative volume of the phoneme string information at an arbitrary point in time. Data
Based on the character string information of the speaking style data, speaking style data storage means for grouping speaking style data having the same character string information but different readings and storing the data for each group, and a word, a phrase or a sentence. Using a character string input means for inputting a character string to be composed, and a character string input through the character string input means, storing a group having the same character string information as the character string in the way of speaking data storage Searching means for searching from the means, voice data storing means for storing a plurality of voice data for adding voice to the synthesized voice,
The speech data in the group retrieved by the retrieval means is sequentially read, and the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speech data and the voice data stored in the voice data storage means are read out. Voice synthesizing means for synthesizing the voice using one of the above, a voice selecting means for selecting a desired voice from the voice synthesized by the voice synthesizing means, and a voice selected using the voice selecting means. Utterance document storage means for storing the corresponding manner of speaking data as an utterance document in association with the character string input via the character string input means.

According to a third aspect of the present invention, in the utterance document creation device according to the first or second aspect, further, a reproduction designation for designating the reproduction of the utterance document stored in the utterance document storage means. Means for reproducing the utterance document, wherein the voice synthesizing means sequentially reads the speaking style data in the utterance document and synthesizes the voice.

According to a fourth aspect of the present invention, in the utterance document creating apparatus according to the third aspect, the reproduction designating means sets an arbitrary character in the utterance document as a range in which the utterance document is reproduced. It is possible to specify a column unit, a sentence unit, a page unit, or the entire utterance document.

According to a fifth aspect of the present invention, there is provided an utterance document creating apparatus, comprising: character string information including words, phrases or sentences; phoneme string information including phonemes corresponding to characters of the character string information; The speech style data comprising the duration of a phoneme, pitch information specifying a relative pitch of the phoneme string information at an arbitrary point in time, and velocity information specifying the volume of each phoneme in the phoneme string information, Based on the character string information of the speaking data, the speaking data data storing means for grouping the speaking data having the same character string information and different readings and storing the data for each group, and forming a word, a phrase or a sentence. A character string input unit for inputting a character string to be input, and a group having the same character string information as the character string using the character string input via the character string input unit. A search means for retrieving from the data storage means, a voice data storage means for storing a plurality of voice data for adding a voice to the synthesized voice, and one of the voice data stored in the voice data storage means. Voice data designating means for reading the voice data in the group searched by the searching means, and sequentially read phonemic sequence information, duration time, pitch information and velocity information of the read voice data and the voice data. A voice synthesizing unit for synthesizing a voice using the voice color data specified by the specifying unit; a voice selecting unit for selecting a desired voice from the voice synthesized by the voice synthesizing unit; The speech style data and the timbre data corresponding to the selected voice are associated with the character string input through the character string input means as an utterance document. And spoken documents storage means for 憶, those having a.

According to a sixth aspect of the present invention, there is provided an utterance document creating apparatus, comprising: character string information comprising words, phrases or sentences; phoneme string information comprising phonemes corresponding to characters of the character string information; Speaking composed of a duration of a phoneme, pitch information specifying a relative pitch of the phoneme string information at an arbitrary point in time, and velocity information specifying a relative volume of the phoneme string information at an arbitrary point in time. Data
Based on the character string information of the speaking style data, speaking style data storage means for grouping speaking style data having the same character string information but different readings and storing the data for each group, and a word, a phrase or a sentence. Using a character string input means for inputting a character string to be composed, and a character string input through the character string input means, storing a group having the same character string information as the character string in the way of speaking data storage Searching means for searching from the means, voice data storing means for storing a plurality of voice data for adding voice to the synthesized voice,
Voice data designating means for designating one of the voice data stored in the voice data storage means; and speaking data in the group searched by the searching means are sequentially read out, and the read voice data is read out. Voice synthesizing means for synthesizing voice using phoneme string information, duration time, pitch information and velocity information and voice data specified by the voice data specifying means, and a desired voice from the voice synthesized by the voice synthesizing means. Voice selecting means for selecting a voice, and speaking data and voice color data corresponding to the voice selected using the voice selecting means are associated with a character string input via the character string input means, and uttered. Utterance document storage means for storing as a document.

According to a seventh aspect of the present invention, in the utterance document creating apparatus according to the fifth or sixth aspect, the utterance document creating apparatus further creates the speaking manner data and registers the speaking manner data in the speaking manner data storage means. It is provided with data creation / registration means.

According to an eighth aspect of the present invention, in the utterance document creating apparatus according to the seventh aspect, the speaking style data creating / registering means is configured to record voice waveform data recorded in advance or a real voice spoken by a user. Voice waveform data input means for inputting and displaying voice waveform data; duration input means for inputting by a user or analyzing the phonetic waveform of the voice waveform data to set and display the duration of each phoneme; Phoneme string information adding means for adding phoneme string information corresponding to the set duration, pitch curve display means for pitch-analyzing the speech waveform data and displaying a pitch curve; A pitch information generator that adjusts or adds a relative pitch value at an arbitrary point in the phoneme string information based on the phoneme string information to generate pitch information; Means, velocity information generating means for adjusting the volume of each phoneme in the phoneme string information to generate velocity information, and character string information setting for inputting a character string corresponding to the speech waveform data and setting character string information Registering means for registering the character string information, phoneme string information, duration, pitch information and velocity information as speaking data in a corresponding group in the speaking data storage based on the character string information And

According to a ninth aspect of the present invention, in the utterance document creation device according to any one of the fifth to eighth aspects, the utterance document creation device further reproduces the utterance document stored in the utterance document storage means. A playback specification means for specifying
When the reproduction of the utterance document is designated, the voice synthesizing means sequentially reads the speech style data and the timbre data in the utterance document and synthesizes the voice.

According to a tenth aspect of the present invention, in the utterance document creating apparatus according to the ninth aspect, the reproduction specifying means sets an arbitrary character in the utterance document as a range in which the utterance document is reproduced. It is possible to specify a column unit, a sentence unit, a page unit, or the entire utterance document.

An utterance document creating apparatus according to claim 11 is the utterance document creating apparatus according to any one of claims 5 to 10, further comprising:
Displaying an utterance document stored in the utterance document storage means, specifying an arbitrary character string of the displayed utterance document, and changing or re-inputting the specified character string; The search is performed by the searching means, the voice data is specified by the voice data specifying means, the voice is synthesized by the voice synthesizing means, and the voice is selected by the voice selecting means, using the character string thus specified, whereby the specified voice is specified. It is possible to change the way of speaking data and voice data corresponding to the character string.

According to a twelfth aspect of the present invention, in the utterance document creating apparatus according to the first, second, fifth or sixth aspect, the plurality of timbre data is a male voice, a female voice, or a child voice. The voice data is sensibly identifiable, such as voice, old man's voice, sharp voice, clear voice, thick voice, thin voice, powerful voice, gentle voice, and mechanical voice.

According to a thirteenth aspect of the present invention, in the utterance document creator according to the first, second, fifth, or sixth aspect, the character string input means has a kana-kanji conversion function, and Example The character string input by the input means is a text with kanji and kana mixed after being converted using the kana-kanji conversion function.

According to a fourteenth aspect of the present invention, in the utterance document creation device according to the fifth or sixth aspect, the utterance document creation device further includes a classification type designating means for designating a classification type of the speaking style data. Method data has type information indicating a classification type of each of the speaking style data in addition to the character string information, phoneme sequence information, duration time, pitch information and velocity information, and When a classification type is designated, the search means uses the character string input via the character string input means and the classification type designated via the classification type designation means to generate the same character string as the character string. Retrieving, from the speaking data storage means, speaking data in a group having character string information and having the same type information as the designated classification type; The speech synthesis means reads the speaking style data retrieved by the retrieval means, and reads the phonetic sequence information, duration time, pitch information and velocity information of the read speaking style data, and the voice specified by the voice color data specifying means. The speech is synthesized using the data.

According to a fifteenth aspect of the present invention, in the utterance document creator according to the fourteenth aspect, the classification type is such that a speech corresponding to each of the speaking style data is a Tokyo type pronunciation type or an Osaka type pronunciation. It is classified by region, such as pronunciation type and Tokushima type pronunciation type.

[0027] According to a twelfth aspect of the present invention, in the utterance document creator according to the fourteenth aspect, the classification type is such that a speech corresponding to each of the speaking style data is an elderly-type pronunciation type and a youth-type. They are classified by age, such as pronunciation type and high school student type.

According to a twelfth aspect of the present invention, in the utterance document creation device according to any one of the fifth to sixteenth aspects, the character string input means has a display unit; Based on the voice data specified for each character string of the utterance document, the font of the character string to be displayed and the decoration method are changed and displayed on the display unit.

According to a twelfth aspect of the present invention, there is provided an utterance document creating method, comprising the steps of: inputting a character string constituting a word, a phrase or a sentence;
Phoneme string information composed of phonemes corresponding to the characters of the character string information, duration time of each phoneme in the phoneme string information, pitch information specifying a relative pitch of the phoneme string information at an arbitrary point in time, and Speaking data composed of velocity information specifying the volume of each phoneme in the phonemic sequence information is converted into speech data having the same character string information and different readings based on the character string information of the speaking data. By referring to the database that has been grouped and stored for each group,
A second step of searching for a group having the same character string information as the character string input in the first step, a third step of specifying voice data to add voice to the synthesized voice, The speech style data in the group searched in the second step is sequentially read, and the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speech style data, and the voice specified in the third step are read out. A fourth step of synthesizing the voice using the data, a fifth step of selecting a desired voice from the voice synthesized in the fourth step, and a step corresponding to the voice selected in the fifth step. And a sixth step of storing the speaking style data to be associated with the character string input in the first step as an utterance document.

According to a twelfth aspect of the present invention, in the utterance document creating method, a first step of inputting a character string constituting a word, a phrase or a sentence, and a character string information comprising a word, a phrase or a sentence,
Phoneme string information composed of phonemes corresponding to the characters of the character string information, duration time of each phoneme in the phoneme string information, pitch information specifying a relative pitch of the phoneme string information at an arbitrary point in time, and Based on the character string information of the speaking data, the speaking style data composed of the velocity information specifying the relative volume at any time of the phonemic sequence information is read. A second step of retrieving a group having the same character string information as the character string input in the first step with reference to a database in which different ways of speaking data are grouped and stored for each group; A third step of specifying voice data to add a voice to the voice to be spoken, and sequentially reading the talking style data in the group searched in the second step, and selecting the voice of the read talking style data. A fourth step of synthesizing the voice using the column information, the duration time, the pitch information and the velocity information, and the timbre data specified in the third step; and a voice synthesized in the fourth step. A fifth step of selecting a desired voice from the first step; and a sixth step of storing the speaking style data corresponding to the voice selected in the fifth step as an utterance document in association with the character string input in the first step. And a process.

According to a twentieth aspect of the present invention, there is provided the utterance document creating method according to the twelfth aspect, further comprising a seventh step of designating reproduction of the utterance document stored in the sixth step. And an eighth step of, when the reproduction of the utterance document is designated in the seventh step, sequentially reading the speech style data and the timbre data in the utterance document and synthesizing the voice.

According to a twelfth aspect of the present invention, in the utterance document creating method according to the twentieth aspect, in the seventh step, an arbitrary range in the utterance document is set as a range in which the utterance document is reproduced. It is possible to specify a character string unit, a sentence unit, a page unit, or the entire utterance document.

According to a twelfth aspect of the present invention, in the utterance document creating method according to any one of the eighteenth to twenty-first aspects, the utterance document stored in the sixth step is further displayed. A ninth step of specifying an arbitrary character string of the displayed utterance document and changing or re-inputting the specified character string, using the character string changed or re-input in the ninth step. By executing the second step, the third step, the fourth step, the fifth step, and the sixth step again, the utterance document can be changed.

Also, a computer-readable recording medium according to claim 23, wherein the utterance document creating method comprises: a first step of inputting a word, a phrase or a character string constituting a sentence;
Character string information composed of words, phrases or sentences, phoneme string information composed of phonemes corresponding to the characters of the character string information, duration of each phoneme in the phoneme string information, Based on pitch information specifying relative pitch and velocity information specifying the volume of each phoneme in the phoneme string information, the character string information is converted into character string information based on the character string information of the talking method data. A second step of searching for a group having the same character string information as the character string input in the first procedure, with reference to a database in which the same but different readings of the speaking data are grouped and stored for each group. Step 2, a third step of designating voice data to add voice to the synthesized voice, and reading the speaking style data in the group searched in the second step sequentially. Out was talking how phoneme sequence information of the data,
A fourth procedure of synthesizing speech using the duration time, pitch information and velocity information, and the timbre data specified in the third procedure, and a desired procedure from the speech synthesized in the fourth procedure. A fifth procedure of selecting a voice, a sixth procedure of storing speaking style data corresponding to the voice selected in the fifth procedure as an utterance document in association with the character string input in the first procedure, And causing a computer to execute a program including the program.

A computer-readable recording medium according to claim 24, further comprising: a first step of inputting a character string constituting a word, a phrase, or a sentence; character string information comprising a word, a phrase, or a sentence; Phoneme string information composed of phonemes corresponding to the characters of the string information, duration time of each phoneme in the phoneme string information, pitch information specifying a relative pitch of the phoneme string information at an arbitrary time, and the phoneme string Based on the character string information of the speaking method data, speaking style data composed of velocity information specifying relative volume at any point in time of the information is spoken with the same character string information and different readings. A second procedure for retrieving a group having the same character string information as the character string input in the first procedure by referring to a database in which the data is grouped and stored for each group A third procedure for specifying voice data for adding a voice to a voice to be synthesized; and a method of sequentially reading the speaking data in the group searched in the second procedure, and phonemic sequence information of the read speaking data. A fourth procedure for synthesizing a voice using the timbre data specified in the third procedure, and a desired voice from the voice synthesized in the fourth procedure. A fifth procedure of selecting the voice of the first step, and a sixth procedure of storing the speaking style data corresponding to the voice selected in the fifth procedure as the utterance document in association with the character string input in the first procedure. Are executed by a computer.

The computer-readable recording medium according to claim 25 is the computer-readable recording medium according to claim 23 or 24, further comprising:
A seventh procedure for designating the reproduction of the utterance document stored in the sixth procedure, and when the reproduction of the utterance document is designated in the seventh procedure, the speech style data and the timbre data in the utterance document are converted. An eighth procedure of sequentially reading and synthesizing speech, and causing a computer to execute the program.

A computer-readable recording medium according to a twenty-sixth aspect is the computer-readable recording medium according to the twenty-fifth aspect, wherein in the seventh step, the utterance document is set as a range in which the utterance document is reproduced. It is possible to specify an arbitrary character string unit, a sentence unit, a page unit, or the entire utterance document in it.

The computer-readable recording medium according to claim 27 is the computer-readable recording medium according to any one of claims 23 to 26, further comprising: the utterance document stored in the sixth procedure. Ninth step of displaying, specifying an arbitrary character string of the displayed utterance document, and changing or re-inputting the specified character string, wherein the character changed or re-input in the ninth step is included. By executing the second procedure, the third procedure, the fourth procedure, the fifth procedure, and the sixth procedure again using the columns, the speech document can be changed. .

[0039]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an utterance document creating apparatus, an utterance document creating method, and a computer-readable recording medium storing a program for causing a computer to execute the utterance document creating procedure according to the present invention will be described (Embodiment 1). , [Embodiment 2] will be described in detail with reference to the accompanying drawings.

Embodiment 1 FIG. 1 is a schematic block diagram of an utterance document creating apparatus 100 according to Embodiment 1. The utterance document creation device 100 includes a control unit 101, an application storage unit 102, a speaking style data storage unit 103, a timbre data storage unit 104,
Voice synthesizing section 105, key input section 106, display section 10
7, a microphone 108, a speaker 109, an utterance document storage unit 110, an interface (I / F) 111, a floppy disk drive (FD drive) 112,
It comprises a CD-ROM drive 113 and a communication unit 114.

The control unit 101 is a central processing unit for controlling the above-mentioned units connected to the bus BS.
1a, a ROM 101b, a RAM 101c, and the like. The CPU 101a stores the O stored in the ROM 101b.
It operates according to an S (operating system) program or an application program stored in the application storage unit 102. The ROM 101b has O
A memory for storing the S program;
Is a memory used as a work area for various programs.

The application storage unit 102 stores an utterance document creation program, an utterance document reproduction program,
Various applications such as a speaking data creation / registration program are stored. Further, the utterance document creating apparatus 100 according to the first embodiment has a kana-kanji conversion function.
The kana-kanji conversion application that realizes the kana-kanji conversion function is also stored in the application storage unit 102.

The speaking style data storage unit 103 plays the role of the speaking style data storage means of the present invention, and, based on the character string information 202 which is one of the information of the speaking style data 201, as shown in FIG. , The character string information 202 groups the same way of speaking data 201,
Is stored in such a manner that a search can be performed on a group basis by using.

The speaking data 201 includes character string information 202 composed of words, phrases or sentences, and character string information 202.
Phoneme string information 203 composed of phonemes corresponding to the characters, the duration time 204 of each phoneme in the phoneme string information 203, pitch information 205 specifying the relative pitch of the phoneme string information 203 at an arbitrary time, a phoneme string The information 203 has velocity information 206 specifying the volume of each phoneme and type information 207 indicating a classification type of each speech style data. Also,
Although detailed description is omitted, it is also possible to search for desired manner of speaking data 201 using information other than character string information 202 (for example, phoneme string information 203 and type information 207) as search keys.

[0045] In this case, for example, when the character string information 202 to search for a group of "Hello", phoneme sequence information 203
Are three types of speaking data 20 of "ko, n, ni, chi, wa" 20
1 is obtained. These speech data 201 have character string information 202 and phoneme string information 203 in common,
When any one of the duration time 204, the pitch information 205, and the velocity information 206 is different, it can be distinguished as different speaking data 201.

For example, when the character string information 202 is searched for the group of "today", the phoneme string information 203 is three types of "ko, n, ni, j, wa" and the phoneme string information 203 is "Kyo, 5 types of talking data 2
01 is obtained. These speech style data 201 can be first distinguished into two pieces by phoneme sequence information 203, and further, any one of the duration time 204, the pitch information 205, and the velocity information 206 is different, so that different speech style data 201 respectively. Can be distinguished as

[0047] In addition, the character string information 202 is "Hello"
The three types of speech data 201 and the character string information 202 of the group
The three types of speaking data 201 of “03,” “ko, n, ni, chi, wa” differ only in character string information 202, and the other information (phonological string information 203 to type information 207) is the same. Therefore, in the first embodiment, for the sake of simplicity, the talking data 201 in the talking data storage unit 103 is shown in the form of a data table as shown in FIG. By dividing the column information 202, the phoneme sequence information 203 to the velocity information 206, and the type information 207, linking them and storing the same information in the form of a common database, Obviously, the overall information can be reduced and the memory can be used efficiently.

The voice data storage section 104 serves as voice data storage means of the present invention, and stores a plurality of voice data for adding voice to a synthesized voice. Here, the timbre data is, for example, spectral information for a phoneme sequence (information that changes over time, and is specifically expressed by cepstrum, LSP parameters, and the like).
Is stored in the form of Further, as shown in FIG. 3, a plurality of voice data includes a male voice, a female voice, a child voice, an old man's voice, a sharp voice, a clear voice, a fat voice, a thin voice,
Voice data that can be discerned intuitively, such as a powerful voice, a gentle voice, and a mechanical voice, is stored.

The speech synthesis unit 105 plays the role of the speech synthesis means of the present invention, and sequentially reads the speech style data 201 in the group of the speech style data storage unit 103 searched by the control unit 101, and reads the read speech style data. The speech is synthesized using the phoneme sequence information 203, the duration time 204, the pitch information 205, the velocity information 206, and one of the timbre data stored in the timbre data storage unit 104.

The key input unit 106 includes input devices such as a keyboard and a mouse, and performs various operations such as input of a character string, selection of voice, designation of reproduction of an utterance document, creation and registration of an utterance document. Used for

The display unit 107 is a liquid crystal display or a CR.
It is composed of a T display and is used for displaying character strings, displaying utterance documents, displaying various messages, and the like.

The microphone 108 is used to sample the original real voice, which is the original voice waveform data, when creating and registering the speaking style data 201.

The speaker 109 is used for reproducing and outputting the voice synthesized by the voice synthesizing unit 105 and reproducing various sounds.

The utterance document storage unit 110 is a memory for storing the prepared utterance document. Although the details will be described later, the utterance document is a document created by associating the selected speech style data 201 and the voice selection number designating the voice data with the character string input via the key input unit 106. is there.

The I / F 111 is a unit for exchanging data between the bus BS and the FD drive 112 or the CD-ROM drive 113. The FD drive 112 carries out an operation of reading and writing data by mounting a removable FD 112a (recording medium). The CD-ROM drive 113 carries out an operation of reading data by mounting a removable CD-ROM 113a (recording medium). The utterance document stored in the utterance document storage unit 110 is stored in the F / F 111 via the I / F 111 and the FD drive 112.
It is also possible to save it in D112a.

The communication unit 114 is connected to a communication line and communicates with an external device via the communication line.

In the first embodiment, the control unit 101,
The key input unit 106 and the display unit 107 serve as character string input means and reproduction designating means of the present invention, and the control unit 1
01 plays the role of search means of the present invention, and the speaker 10
9. The key input unit 106 and the control unit 101 play the role of voice selection means and voice data designation means of the present invention,
The control unit 101 and the utterance document storage unit 110 function as the utterance document storage unit of the present invention, and the control unit 101, the key input unit 106, the display unit 107, the microphone 108, and the speaker 109 generate and register the speaking style data of the present invention. Act as a means.

In the first embodiment, the key input unit 10
A description will be given of a case where a character string is input through the input device 6 as an example. However, the present invention is not particularly limited to this, and a handwriting input device is connected to determine a handwritten character (character recognition) and input the character string. Alternatively, a character string may be input from a word processing document or the like created in advance.

FIG. 4 is an external view of the utterance document creating apparatus 100 according to the first embodiment. As shown in the figure, as a hardware configuration, a personal computer having a microphone 108 and a speaker 109 can be used.

In the above configuration, a speech document creation process a speech document playback process a speech document creation process using type information a speech document playback process using type information creation / registration process of speaking style data Changing a speech document The operation will be described in the order of processing.

Speech Document Creation Process The speech document creation process will be described with reference to FIGS. Here, FIG. 5 shows a schematic flowchart of the creation process of the utterance document, and FIGS. 6 to 9 show examples of the display screen of the display unit 107 in the creation process of the utterance document. Here, when the main body of the utterance document creation apparatus 100 is turned on, the control unit 101 starts the utterance document creation program stored in the application storage unit 102, and
Is executed.

First, the document creator uses the key input unit 106 and the display unit 107 to input a word, a phrase, or a character string constituting a sentence (S501: a first step of claim 18). For example, through the key input unit 106 inputs a character string "Hello", the display unit 107, FIG. 6
As shown on the display screen D1 of (a), the character string "Hello" is displayed. Although it is possible to accept the string "Hello", wherein further using the kana-kanji conversion function, the display screen D shown in FIG. 6 (b)
As shown in 2, it is assumed that the use of the "Hello" → text that was converted to a string of kanji kana and Pepper of "Today."

Next, a group having the same character string information 202 as the character string "today" entered in step S501 is searched from the speaking data storage unit 103 (S50).
2: The second step of claim 18). In other words, the speech data 201 corresponding to the character string “today” is searched.
Specifically, as shown in FIG. 2, the speaking style data storage unit 103 stores the character string information 2 corresponding to the character string “today”.
02, the phoneme sequence information 203 is “Ko, N,
D, J, W ”and the phoneme sequence information 203 are“ Kyo,
5), a total of five types of speaking manner data 201 of "U, W".

Subsequently, designation of voice data is selected, and voice data for adding a voice to the voice to be synthesized is specified (S503, S504: a third step of claim 18). Specifically, for example, a voice color designation button 701 is displayed as shown in a display screen D3 in FIG. 7A, and this button is clicked with a mouse, and as shown in a display screen D4 in FIG. , Can be realized by displaying and selecting voice data stored in the voice data storage unit 104. Here, the voice selection number (the number corresponding to the voice data in FIG. 7B) corresponding to the selected voice data is held.
Thereafter, voice data is specified by the voice selection number. If designation of voice data is not selected, it is assumed that the previously specified voice data (that is, the previously selected voice selection number) has been specified again, and step S505 is performed.
Shall proceed to

Next, the speech synthesizing unit 105 determines in step S5
02 is read out sequentially in the group searched in 02, and the phoneme sequence information 203, the duration time 204, the pitch information 205 and the velocity information 206 of the read out speech style data 201 and the designated voice data are used. Then, the voice is synthesized, and the synthesized voice is output via the speaker 109 (S505: fourth step of claim 18). Specifically, the phoneme string information 203 belonging to the searched group is three kinds of speaking data 201 of “ko, n, ni, chi, wa” and the phoneme string information 203 is two of “kyo, u, wa”.
The type of speech data 201 is sequentially synthesized and output.

The document creator listens to the sequentially reproduced speech style data 201 and selects a desired voice (S506:
(Fifth step of claim 18). Here, steps S505 to S506 are repeated until a desired voice is selected.

When the desired voice is selected in step S506, the voice data at that time (voice selection number), the speaking data 201 corresponding to the selected voice, and the step S5
The utterance document data is created by associating the character string “today is” with the character string “01” input in step S 01, stored in the utterance document storage unit 110 (S 507: sixth step of claim 18), and a predetermined end key is designated. Steps S501 to S507 are repeated until this is done (S508).

FIG. 8 shows an example of a screen display of the utterance document created in the above-mentioned utterance document creation processing. FIG. 9 shows an example of the utterance document data stored in the utterance document storage unit 110.
In the utterance document shown in FIG. 8, for example, a female voice is specified as the voice data of the part of Hanako-san's words "Today is Ichitaro-san" or "No, patent drawing ...". By specifying a male voice as voice data of a portion other than the above, it is possible to reproduce one utterance document with a plurality of voice data at the time of reproduction via the voice synthesis unit 105.

Also, the character string "today is" 8
The character string “today” indicated by 02 is the same as the character string on the screen display,
02, the phoneme sequence information 2 of the speaking style data 201
03 is different, the character string 801 is (ko, n, d, j,
And the character string 802 is pronounced as (Kyo, U, Wa). Therefore, the document can be accurately vocalized in the reading manner intended by the document creator.

As described above, in the process of creating the utterance document, the utterance document data in which the input character string (character information) is matched with the voice information (speech method data) including the way of speech intended by the document creator. Can be created.

The utterance document data (in other words, the way of speaking data 201) includes the duration time 204, the pitch information 205, and the velocity information 20 in addition to the phoneme sequence information 203.
6, and the document creator can actually generate speech document data by actually listening to the speech synthesized from the speech data 201, so that accent, loudness (volume),
By adjusting the pitch of the voice and the like to create information (speech document data) having the intended way of speaking, it is possible to add an expression such as emotion corresponding to non-verbal information as information.

As the expression of emotions, for example, the speech synthesis data 201 of the character string “OK” is synthesized by voice, and the following two types are selected to express the emotions intended by the document creator. be able to.

1) When the character string "I understand" is spoken in a rising tone, the sentence becomes a question asking whether or not it is understood. Etc. can be included. 2) Change the character string "I understand" to "ta"
When the volume of only the part is increased and the speaking is short, while conveying the understanding or acceptance in the literal sense as information, while using in the text,
It is possible to express as a feeling that one fully understands, understands, but is uncomfortable, that he / she is unwilling to accept.

Reproduction Process of Utterance Document Next, the reproduction process of the utterance document will be described with reference to FIGS. Here, FIG. 10 shows a schematic flowchart of the reproduction process of the utterance document, and FIGS. 11 to 13 show examples of the display screen of the display unit 107 in the reproduction process of the utterance document. Although illustration is omitted, when the reproduction process of the utterance document is selected on the display screen of the display unit 107, the control unit 10
1 activates the utterance document reproduction program stored in the application storage unit 102, and executes the schematic flowchart of FIG.

First, a list of utterance documents stored in the utterance document storage unit 110 is displayed on the display unit 107 to prompt the document creator to select an utterance document to be reproduced. The document creator selects an utterance document via the key input unit 106 (S100)
1: the seventh step of claim 20), the selected utterance document is read out from the utterance document storage unit 110, and displayed on the display unit 107.
(S1002). At this time, as shown in FIG. 11, based on the voice data specified for each character string of the utterance document, the font of the character string to be displayed and the decoration method (for example, halftone dots, reverse display, etc.) are changed. It is convenient to be able to visually recognize the difference in the timbre data by displaying the data.

Next, the key input unit 106 and the display unit 10
The reproduction range of reproducing the utterance document by using 7 is shown in the display screen of FIG. ) (S1003). Here, for example, if the character string unit of 1) is selected and an arbitrary character string unit (at least one character string) in the utterance document is specified as shown in the display screen of FIG. The displayed character string 1301 is highlighted. When the text unit of 2) is selected and an arbitrary text unit (at least one text unit) in the utterance document is specified as shown in the display screen of FIG. 13B, the specified text 1302 is inverted. Is displayed. When the page unit of 3) and the entire utterance document of 4) are designated, the screen display remains the same as in FIG. 11 and the designated page number or a message indicating the designation of the entire document is displayed.

When the reproduction range is specified in step S1003, the speech synthesis unit 105 sequentially reads out the corresponding utterance document data (speaking style data and voice color data) in the utterance document based on the specified reproduction range. (S1004: Eighth step of the twentieth aspect).

After that, when the voice synthesis in the designated reproduction range is completed, step S100 is performed until a predetermined reproduction processing end button (not shown) on the display unit 107 is selected.
Steps S3 to S1004 are repeated (S1005).

As described above, in the reproduction process of the utterance document,
Since the utterance document is created in advance as utterance document data in which the character string (character information) and the voice information (speech data) including the way of speech intended by the document creator are created, the utterance document is displayed on the display screen. Only the voice to be reproduced can be visually selected from the displayed utterance document (displayed character string).

Further, the utterance document data (in other words, the way of speaking data 201) includes the duration time 204, the pitch information 205, and the velocity information 2 in addition to the phoneme sequence information 203.
Since the document creator actually listens to and creates a speech obtained by speech-synthesizing the speech style data 201, the speech is reproduced as a speech to which an expression such as emotion corresponding to non-verbal information is added. be able to.

A process for creating an utterance document using type information The process for creating an utterance document using type information will be described. FIG. 14 is a schematic flowchart of a process of creating an utterance document using type information. Although not shown, when the process of creating an utterance document using type information is selected on the display screen of the display unit 107, the control unit It is assumed that 101 activates an utterance document creation program using the type information stored in the application storage unit 102, and executes the schematic flowchart of FIG.

Since the schematic flowchart of FIG. 14 is basically the same as the schematic flowchart of the speech document creating process shown in FIG. 5, the same steps are denoted by the same reference numerals, and different parts are used here. Only the explanation will be given.

First, the key input unit 106 and the display unit 10
7 is used to specify the classification type of the speaking style data (S
1401). Here, the classification type refers to, for example, a speech corresponding to each type of speech data, which is classified according to a region such as a Tokyo-type pronunciation type, an Osaka-type pronunciation type, a Tokushima-type pronunciation type, and the like, and an elderly-type pronunciation type. , Younger type pronunciation type, high school student type, etc., classified by age can be used. In other words, a classification type is determined in advance, and based on the classification type, for example, in the case of Osaka type pronunciation type, Kansai-style speaking style data 201 is created and classified as Osaka type pronunciation type. Type information 20 of the speaking data 201
7

FIG. 15 shows an example of a classification type designation screen. Here, TYPE1: Tokyo type, TYPE2: Osaka type, TYPE3: Elderly type, TYPE4: Youth type, TYPE
E5: It is assumed that five classification types of high school student type are prepared in advance.

After the classification type is specified, a character string is input (S5
01) is performed, then the talking data storage unit 103
From the character string input in step S501 and the designated classification type, a group having the same character string information as the character string and a speech having the same type information as the designated classification type The speech data 201 is retrieved from the speaking data storage unit 103 (S1402). In other words, only the way of speaking data 201 of the corresponding classification type is searched. In this case, the speaking data 201 of the corresponding classification type is stored in the speaking data storage unit 103.
When there are a plurality of utterances, a plurality of talking data 201 is searched.

Subsequently, designation of voice data (S503, S
504) is executed.

Next, the speech synthesizing unit 105 determines in step S1
The speech data 201 retrieved in 402 is read out, and the speech is generated using the phoneme sequence information 203, the duration time 204, the pitch information 205, and the velocity information 206 of the read speech data 201, and the designated timbre data. The synthesized voice is output through the speaker 109 (S505). Here, since the classification type is specified, only the corresponding speech data 201 is synthesized.

Thereafter, when a desired sound is selected (S50)
6), utterance document data is created and utterance document storage unit 110
(S507), and steps S1401, S501, S1402, and S502 to S507 are repeated until a predetermined end key is designated (S508). Note that 2
In the subsequent steps S1401, it is assumed that the process immediately proceeds to step S501 to input a character string unless the classification type is changed.

As described above, in the process of creating an utterance document using type information, the classification type of the speaking style data 201 to be reproduced by performing speech synthesis can be designated. Can be easily created, which is convenient. In addition, it is possible to reduce the time required for creating the utterance document.

In the flowchart of FIG. 14, every time a character string is input, steps S503 to S50 are performed.
6, the voice color data is specified and the voice is selected. However, the present invention is not limited to this. When the method 201 searches for the appropriate type of speaking data 201 in step S1402, the process proceeds to step S507. Alternatively, the utterance document may be automatically stored using the searched speech style data 201. In this case, after a character string constituting the utterance document is input to some extent, the process proceeds to step S5.
03 to step S506 are executed so that voice data can be specified for each character string.

Reproduction processing of utterance document using type information Reproduction processing of an utterance document using type information specifies a classification type to be used for reproduction, and specifies the classification type and the specified classification type.
Using the character string information 202 and the phoneme string information 203 in the utterance document (utterance document stored in the utterance document storage unit 110) created in the utterance document creation process of the above, the corresponding speech data storage unit 103 The speech data 201 to be spoken is searched, and voice data in an utterance document (an utterance document stored in the utterance document storage unit 110) created in the utterance document creation process with the searched speech data 201 is used. , A voice synthesizer 105 synthesizes a voice, and a speaker 109
And reproduces the output.

In other words, without using the duration 204, pitch information 205 and velocity information 206 of the speaking style data 201 specified in the process of creating the utterance document,
Speaking data 20 specified by type information 207
1 uses the duration 204, pitch information 205, and velocity information 206.

FIG. 16 is a schematic flowchart of the reproduction process of the utterance document using the type information. Although not shown, when the reproduction process of the utterance document using the type information is selected on the display screen of the display unit 107. The control unit 101 activates an utterance document reproduction program using the type information stored in the application storage unit 102, and executes the schematic flowchart of FIG.

First, a list of the utterance documents stored in the utterance document storage unit 110 is displayed on the display unit 107 to prompt the document creator to select an utterance document to be reproduced. The document creator selects an utterance document via the key input unit 106 (S160)
1), the selected utterance document is read from the utterance document storage unit 110 and displayed on the display unit 107 (S160).
2).

Next, the key input unit 106 and the display unit 10
7 to specify the classification type to be used for reproduction (S
1603). The classification type can be specified using the display screen of FIG.

Subsequently, using the designated classification type and the character string information 202 and the phoneme string information 203 in the selected utterance document, the corresponding speaking data 201 is searched from the speaking data storage unit 103. (S1604).

Next, the speech synthesizing unit 105 generates the phoneme sequence information 203 of the searched speech style data 201 and the continuation time length 20.
4, pitch information 205 and velocity information 206,
Voice data in the selected utterance document (voice data of utterance document data including phoneme string information 203 used for search)
Are used to synthesize and reproduce and output through the speaker 109 (S1605). As a result, the character string information 2 corresponding to the specified classification type and voice data can be obtained.
02 and the phoneme string information 203 are speech-synthesized.

Thereafter, it is determined whether or not all the character strings of the selected utterance document have been voice-synthesized (S1606), and steps S1604 and S1605 until all the character strings in the corresponding utterance document are voice-synthesized and output. Is repeated, and the process ends when the output is completed.

By executing the reproduction process of the utterance document using the type information as described above, even if the utterance document which has already been created has a way of speaking (that is, the way of speaking data 201) set, the classification type By specifying
It is possible to play sound in different ways of speaking.

Next, a method for creating new speaking data 201 and registering it in the speaking data storage unit 103 will be described with reference to FIGS. 17 to 25. The speech data 201 is shown in FIG.
As shown by, the character string information 202 and the phoneme string information 20
3, a duration length 204, pitch information 205, velocity information 206, and type information 207. Therefore, the creation of the speaking style data 201 is to create or set such information.

It is assumed that, as a rule, a plurality of speaking style data 201 are prepared in advance as standard data and registered in the speaking style data storage unit 103, but freely according to the individual sense of the document creator. Speaking data 20
By creating and registering 1, the choice of how to speak (voice) can be expanded, and the expressiveness of the utterance document can be further increased.

FIG. 17 is a schematic flowchart of the process of creating and registering the speaking style data. First, a pre-recorded voice waveform data is input or a natural voice (a real voice spoken by the user) is input using the microphone 108 (S170).
1) After analyzing and digitizing the input natural speech,
The audio waveform data is generated and displayed on the display unit 107 (S
1702). Here, the pre-recorded voice waveform data means that the natural voice is input in advance using the microphone 108 in the utterance document creating apparatus 100, and is input via the application storage unit 102, the I / F 111 and the FD drive 112. The audio waveform data stored in the FD 112a is shown. Also, audio waveform data recorded by another device may be input and used.

The generated speech waveform data is shown in FIG.
It is displayed on the display screen of the display unit 107 as shown by 0B. FIG. 18 shows a display screen of the creation / registration processing of the speaking style data displayed on the display unit 107, and the syllable display window 10 which is a window for displaying the phoneme sequence information 203.
A, an original waveform display window 10B which is a window for displaying audio waveform data generated from the input natural audio,
A synthesized waveform display window 10C which is a window for displaying waveform data synthesized from the speaking style data 201, a pitch display window 10D which is a window for displaying pitch information 205, and a velocity display window 10E which is a window for displaying velocity information 206. And an original audio playback / stop button 10 for designating audio playback / stop of the audio waveform data displayed in the original waveform display window 10B.
F and sound reproduction / designation for designating sound reproduction / stop of the waveform data displayed in the synthesized waveform display window 10C.
A stop button 10G, a pitch reference setting scale 10H for setting a pitch reference of the pitch information 205, and a character string input area 10Y for inputting character string information 202.
And

Next, the phoneme analysis of the speech waveform data generated in step S1702 is performed to determine the duration of each phoneme, and a label is generated by visualizing the determined duration of each phoneme on the time axis. Display on the display unit 107 (S170
3). Here, the visualized label indicates a line 10I that vertically crosses each of the windows 10A to 10E as shown in the display screen of FIG. The position of each label 10I automatically given by the phonetic analysis is determined by the key input unit 10
Move (or change) manually using mouse 6
Make it possible. This is to allow the label 10I to be assigned to a more appropriate position when the accuracy of the phoneme analysis is low.

Subsequently, phoneme string information corresponding to the space partitioned by the set label (that is, the duration time) 10I is input (S1704). Specifically, a corresponding phoneme (character) is input between the labels 10I of the syllable display window 10A by a manual operation using the key input unit 106. FIG. 20 shows an example in which phoneme sequence information 203 is input, and “yo”, “b”, “Ushi”,
An example is shown in which phonemes are input in the order of "i", "de", "Us", ",", and "ka". Of the input phonemes,
"Ushi" and "Usu" indicate unvoiced phonemes, and the others indicate voiced phonemes.

In the following step S1705, pitch analysis is performed on the speech waveform data to display a pitch curve. FIG. 21 shows a pitch curve after pitch analysis displayed on the pitch display window 10D.

In the following step S1706, pitch adjustment is performed. The pitch adjustment includes operations such as addition and deletion of a pitch label to be described later, and change of a pitch reference pitch value. That is, in step S1706,
The pitch information 205 is generated by adjusting or adding the value of the pitch at an arbitrary point in the phoneme sequence information 203. FIG.
An example is shown in which a pitch label 10J is added in pitch adjustment, and a pitch label 10J is further added to a label 10I that partitions phonemes. This additional operation may be performed by directly specifying the label position in the pitch display window 10D with a mouse or the like. Since the newly added pitch is connected to the adjacent pitch by a straight line, a desired pitch change can be given in one phoneme, and it is easy to process the voice to an ideal voice quality.

In the following step S1707, step S
A synthesized waveform at the stage after the pitch adjustment is completed by the processing up to 1706 is generated, and the synthesized waveform data is displayed in the synthesized waveform display window 10C, for example, as shown in the display screen of FIG. At this time, since the velocity has not been set, a plain velocity is displayed in the velocity display window 10E as shown in the figure.

Although a detailed description is omitted, in step S1707, the synthesized waveform data displayed in the synthesized waveform display window 10C and the original voice waveform data displayed in the original waveform display window 10B are combined. Comparison playback is possible. At this stage, the voice type (voice data) of the synthesized voice is a default voice. More specifically, the user can operate the audio reproduction / stop button 10G to reproduce or stop the synthesized waveform data, and the original audio reproduction / stop button 10F
Can be operated to reproduce or stop the audio waveform data.

In the following step S1708, the velocity (velocity information 206) indicating the volume of the phoneme is manually adjusted. That is, the volume of each phoneme in the phoneme sequence information 203 is adjusted to generate velocity information 206.
This velocity adjustment is performed for each phoneme as shown in FIG. 24, and a predetermined range of steps (for example,
16).

When the reproduction of the synthesized waveform data is operated again after the velocity adjustment, the intensity of the voice changes for each phoneme, and the voice (voice) is inflected as compared with the plain velocity state. Will be added.

Next, in step S1709, a character string corresponding to the audio waveform data intended by the document creator (here, the creator of the way of speaking data) is input, and character string information 2 is entered.
Set 02. For example, as shown in FIG. 25, when a character string “Okay?” Is input to the character string input area 10Y via the key input unit 106, this character string “Okay?” Is set as the character string information 202. .

In the following step S1710, based on the set character string information 202, the speaking style data storage unit 10
The corresponding group of No. 3 is searched, and the talking style data 201 is additionally registered in the searched group. That is, the character string information 202 set in the character string input area 10Y and the phoneme string information 203 input in the syllable display window 10A.
And the duration 20 set as a visualized label
4, the pitch information 205 set in the pitch display window 10D, and the velocity information 206 set in the velocity display window 10E, and the speaking style data 201 is generated and stored in the speaking style data storage unit 103.

Although the description is omitted, it is assumed that the type information 207 of the talking style data 201 registered in this way is set by separately executing classification type setting / change processing after the registration of the talking style data 201. I do. This takes into consideration that if the task of creating the speaking style data 201 and the task of setting the classification type are performed at the same time, the sensation of the document creator becomes dull and the type cannot be classified properly. Add step and type information 20
It goes without saying that the setting of 7 may be performed.

In the first embodiment, natural speech is input using the microphone 108 to generate speech waveform data, but one of the speaking style data 201 stored in the speaking style data storage unit 103 is designated. And input it as original speech waveform data.
The duration time 204, the pitch information 205, and the velocity information 206 are corrected, and the character string information 202 and the phoneme sequence information 203 included in the speech style data 201, and the corrected duration time 204, the pitch information 205, and the velocity information 206 are combined. New talking data 201 may be created by using the data and registered in the talking data storage unit 103.

In the first embodiment, step S17
In step S1704, the label is generated after generating the label in step S1704. However, for example, the label may be generated after inputting the phoneme sequence information. Furthermore, by using the speech recognition technology, it is possible to automate from input of phoneme string information to generation of a label.

Speech Document Change Process The utterance document change process is to display the utterance document stored in the utterance document storage unit 110 again on the display unit 107, and to convert the character string and the speech style data 201 constituting the utterance document. To change.

FIG. 26 shows a schematic flowchart of the process of changing the utterance document. First, a list of utterance documents stored in the utterance document storage unit 110 is displayed on the display unit 107, and the document creator is prompted to select an utterance document to be changed. The document creator selects an utterance document via the key input unit 106 (S2).
601), the selected utterance document is stored in the utterance document storage unit 11
0 and read it out on the display unit 107 (S260).
2).

Next, an item to be changed is designated from a display screen (not shown) (S2603). Here, the items that can be changed include 1) a change in a character string in the utterance document, 2) a change in speaking style data corresponding to the character string, 3) a change in information in the speaking style data, and 4) a change in voice data. There are changes.

Next, when a character string to be changed is specified (S2604), the change item specified in step S2603 is determined (S2605), and the process proceeds to any of steps S2606 to S2609 according to the change item.

1) If the character string in the utterance document is to be changed, the flow advances to step S2606 to execute a character string changing process. This character string change process basically performs the same process as the schematic flowchart of the speech document creation process shown in FIG. The different parts are described in step S in FIG.
507, using the created utterance document (that is, the utterance document created using the input character string), the utterance document of the character string portion designated to be changed (the original utterance document stored in the speaking style data storage unit 103). Utterance document).

2) If the speaking style data corresponding to the character string is to be changed, the flow advances to step S2607 to execute a speaking style data changing process. This process of changing the way of speaking data basically executes other steps except for step S501 from the schematic flowchart of the process of creating the utterance document shown in FIG. The difference is that the step S507 of FIG. 5 uses the created utterance document (that is, the utterance document after the change in the way of speaking data) to store the utterance document of the character string portion designated to be changed (in the manner of speaking data storage unit 103). Original utterance document).

3) If the information in the speaking style data is to be changed, the flow advances to step S2608 to execute processing for changing the information in the speaking style data. The process of changing the information in the speaking style data can be basically realized by the same method as the process of creating and registering the speaking style data shown in FIG. That is,
Character string information 202, phoneme string information 203, and duration 2 of the way of speaking data 201 of the character string portion designated for change
04, pitch information 205 and velocity information 206 as original information, respectively, in the character string input area 1
0Y, syllable display window 10A, visualized label, pitch display window 10D, velocity display window 10
After setting to E, the adjustment of the visualized label, the adjustment of the pitch, and the adjustment of the velocity are performed to change the speaking data 201.

4) In the case of changing the timbre data, the flow advances to step S2609 to execute a timbre data change process. This process of changing the speaking style data basically executes steps S503 and S504 of the schematic flowchart of the process of creating the utterance document shown in FIG.
That is, the voice data in the utterance document (utterance document data) corresponding to the character string whose voice data is to be changed is replaced with the newly specified voice data.

As described above, since the utterance document stored in the utterance document storage unit 110 can be changed, the created utterance document can be used efficiently.
For example, a standard utterance document can be created, and only necessary parts can be changed and used.

As described above, according to the first embodiment, character information (character strings) and voice information (speaking data) are used.
It is possible to create information (an utterance document) in which information of two expression forms is combined in a consistent manner.

Also, in the utterance document created by the utterance document creation device 100, the character information and the speech information (speech data) including the speech style intended by the document creator have a one-to-one correspondence. Even if the same operation as moving or copying a document is performed in a document creation device (Japanese word processor, English word processor), the matching of character information and voice information is not broken, and the spoken document can be easily edited. Therefore, it is possible not only to listen, but also to proceed with the work while watching, thereby making it easy to edit audio information.

Further, according to the purpose of use, both characters and voice can be presented at the same time, and either one of the information can be cut out and only one of the information can be presented. For example, when an utterance document created by the utterance document creation device of the present invention is received by e-mail or the like, it is possible to take out only voice information (speech synthesized using the way of speaking data) from the outside by telephone. is there.

Further, since the document creator can select a desired voice (speech data) to create an utterance document, when performing speech synthesis based on the created utterance document, errors in reading, accents, etc., may occur. In other words, it is possible to output an accurate sound intended by the document creator.

Further, the speech document creation procedure described in the first embodiment can be programmed and stored as a computer-executable program on a computer-readable recording medium.

[Embodiment 2] Embodiment 2 enables editing of speech style data 201 (processing for changing information in speech style data) in the process of creating an utterance document. The velocity information 206 of 201 is information specifying the relative volume of the phoneme sequence information 203 at an arbitrary point in time. In addition,
The basic configuration and operation are the same as those of the utterance document creating apparatus 100 according to the first embodiment, and only different portions will be described here.

FIG. 27 is a schematic flowchart of the process of creating an utterance document according to the second embodiment. The basic operation is the same as that of the utterance document creation process according to the first embodiment shown in FIG. 5, and the same steps are denoted by the same reference numerals and will be briefly described.

First, the document creator uses the key input unit 106 and the display unit 107 to input a word, a phrase, or a character string constituting a sentence (S501: a first step of claim 19). Next, the same character string information 202 as the character string input in step S501 is read from the speaking style data storage unit 103.
(S502: second step of claim 19).

Subsequently, designation of voice data is selected, and voice data for adding a voice to a synthesized voice is specified (S503, S504: a third step of claim 19). Here, the voice color selection number corresponding to the selected voice color data is held, and thereafter, the voice color data is specified by the voice color selection number. If designation of voice data is not selected, it is assumed that the previously specified voice data (that is, the previously selected voice selection number) has been specified again, and the process proceeds to step S505.

Next, the speech synthesizing unit 105 determines in step S5
02 is read out sequentially in the group searched in 02, and the phoneme sequence information 203, the duration time 204, the pitch information 205 and the velocity information 206 of the read out speech style data 201 and the designated voice data are used. Then, the voice is synthesized, and the synthesized voice is output via the speaker 109 (S505: a fifth step of claim 19).

Subsequently, the document creator listens to the sequentially reproduced speech style data 201 and selects a desired voice, or
Alternatively, if the desired voice does not exist, after selecting the editing of the speaking style data, the closest voice is selected. The selection of the editing of the way of speaking data is performed in the same manner as in the display screen for specifying the voice data shown in FIG. Steps S506 and S27 are performed according to the selection.
At 01, it is determined whether a voice has been selected or whether editing of speaking style data has been selected.

When the desired voice is selected, the voice data (voice color selection number) at that time, the speaking style data 201 corresponding to the selected voice, and the character string input in step S501 are made to correspond to each other, and The data is created and stored in the utterance document storage unit 110 (S507: the sixth step of claim 19).
Steps 501 and subsequent steps are repeated (S508).

On the other hand, if the editing of the way of speaking data is selected, the flow advances to step S2702 to determine whether or not the closest voice has been selected. If the closest voice has been selected, the flow advances to step S2703 to be described later. Thus, the schematic flowchart of the information change process in the speaking style data of FIG. 28 is executed.

Thereafter, the speech data 201 changed by the information changing process in the speech data, the voice data at that time (voice selection number), and the character string input in step S501 are made to correspond to each other, and Is created and stored in the utterance document storage unit 110 (S507), and the process from step S501 is repeated until a predetermined end key is designated (S508).

FIG. 28 is a schematic flowchart of a process for changing information in the talking style data according to the second embodiment.
First, the speaking style data 2 corresponding to the selected closest sound
01, the character string information 202, the phoneme string information 203, the duration time 204, the pitch information 205, and the velocity information 206 are read from the speaking style data storage unit 103 (S
2801).

Next, as shown in FIG.
Character string information 202 and phoneme string information 20 read out at 801
3, the duration length 204, the pitch information 205 and the velocity information 206 are respectively entered in the character string input area 10Y,
The syllable display window 10A, the visualized label, the pitch display window 10D, and the velocity display window 10E are set (that is, displayed) (S2802). At this time, the waveform data synthesized from the speaking data 201 is displayed in the original waveform display window 10B.

Subsequently, on the display screen of FIG. 29, the information of the speaking style data 201 is changed by performing the adjustment of the visualized label, the adjustment of the pitch, or the adjustment of the velocity (S2803). In the second embodiment, the velocity information 206 of the speaking style data 201 is replaced with the phoneme sequence information 203.
Can be specified and adjusted as a relative volume at any point in the phoneme sequence information 203 regardless of the unit. Specifically, apart from the label 10I indicating the unit (delimiter) of the phoneme string information 203, the label 10K is set at an arbitrary position.
Can be specified to adjust the volume (velocity information 206). As a result, it is possible to further edit the manner of speaking.

Next, a synthesized waveform is generated based on the adjusted information, and for example, as shown in the display screen of FIG. 30, the synthesized waveform data is displayed in a synthesized waveform display window 10C, and voice synthesis is performed. To reproduce (S2804).
Although a detailed description is omitted, in step S2802, the waveform data synthesized from the synthesized waveform data displayed in the synthesized waveform display window 10C and the original speaking data displayed in the original waveform display window 10B. Can be compared and reproduced.

Thereafter, the processing of steps S2803 to S2804 is repeated until a predetermined end key is designated (S2805).

As described above, according to the second embodiment, it is possible to edit detailed information (adjustment of labels, adjustment of pitch, and adjustment of velocity) in speech style data during creation of an utterance document. Therefore, the convenience can be further improved.

Further, since the velocity information 206 of the speaking style data 201 is the information for designating the relative volume at any time of the phoneme string information 203, the speaking style data intended by the document creator can be easily created. At the same time, it is possible to create more diverse ways of speaking.

[0147]

As described above, the utterance document creating apparatus according to the present invention (claim 1) provides character string information comprising words, phrases or sentences, and phoneme string information comprising phonemes corresponding to the characters of the character string information. Speech, which is composed of the duration of each phoneme in the phoneme sequence information, pitch information specifying the relative pitch of the phoneme sequence information at any time, and velocity information specifying the volume of each phoneme in the phoneme sequence information Based on the character string information of the speaking style data, the speaking style data storing means for grouping speaking style data having the same character string information and different readings and storing the data for each group; Or, using a character string input means for inputting a character string constituting a sentence, and a character string input via the character string input means, storing a group having the same character string information as the character string in a manner of speaking data storage Search means for searching from the first stage, voice data storage means for storing a plurality of voice data for adding voice to synthesized voices, and reading method of speaking in the group retrieved by the searching means sequentially, and the read method of speaking. Speech synthesis means for synthesizing speech using phonological sequence information, duration time, pitch information and velocity information of the data and one of the timbre data stored in the timbre data storage means, and speech synthesized by the speech synthesis means Means for selecting a desired voice from the user, and speaking style data corresponding to the voice selected using the voice selecting means are associated with the character string input through the character string input means as a speech document. Utterance document storage means for storing the information (speech document), which is a combination of two types of information, character information and voice information, in a consistent manner. It can be formed. It also creates information (spoken documents) that matches the text information with the speech information (speech data) including the way the creator intends to speak, and expresses emotions, etc., which correspond to non-verbal information in the document. Can be added. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

Further, the speech document creating apparatus of the present invention (claim 2) is characterized in that character string information composed of words, phrases or sentences, phoneme string information composed of phonemes corresponding to the characters of the character string information, and phoneme string information in the phoneme string information. Speech style composed of duration time of each phoneme, pitch information specifying relative pitch at any time of phoneme sequence information, and velocity information specifying relative volume at any time of phoneme sequence information Based on the character string information of the speaking method data, speaking method data storage means for grouping speaking data having the same character string information and different readings and storing the data for each group; Using a character string input means for inputting a character string constituting a sentence, and a character string input via the character string input means, a group having the same character string information as the character string is used as a data storage method for speaking. Search means for retrieving voice data; voice data storage means for storing a plurality of voice data for adding voice to synthesized voice; and speech data read sequentially from the group retrieved by the search means. Speech synthesis means for synthesizing speech using the phoneme sequence information, duration time, pitch information, and velocity information of the voice and one of the timbre data stored in the timbre data storage means, and a speech synthesized by the speech synthesis means. Voice selecting means for selecting a desired voice, and speaking data corresponding to the voice selected using the voice selecting means are stored as an utterance document in association with the character string input via the character string input means. Utterance document storage means to generate information (speech document) in which information in two representation forms, textual information and voice information, is combined in a consistent manner. It can be. It also creates information (spoken documents) that matches the text information with the speech information (speech data) including the way the creator intends to speak, and expresses emotions, etc., which correspond to non-verbal information in the document. Can be added. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

Further, the speech document creation device of the present invention (claim 3) can designate the reproduction of the speech document, sequentially read the speaking style data in the speech document, and synthesize the speech, so that the speech document can be synthesized. Can be easily confirmed.

Further, the speech document creation device of the present invention (claim 4) can designate an arbitrary character string unit, a sentence unit, a page unit, or the entire speech document in the speech document as a reproduction range of the speech document. Therefore, reproduction and confirmation of the utterance document can be easily performed.

Further, the speech document creating apparatus of the present invention (claim 5) is characterized in that character string information composed of words, phrases or sentences, phoneme string information composed of phonemes corresponding to the characters of the character string information, and phoneme string information in the phoneme string information. Speaking style data consisting of duration time of each phoneme, pitch information specifying relative pitch of phoneme sequence information at an arbitrary point in time, and velocity information specifying volume of each phoneme in phoneme sequence information Based on character string information of the data, the speaking data having the same character string information but different readings are grouped and stored in each of the speaking manner data storing means; and a character constituting a word, a phrase or a sentence. Using a character string input means for inputting a string and a character string input via the character string input means, a group having the same character string information as the character string is retrieved from the speaking data storage means. Searching means; voice data storing means for storing a plurality of voice data for adding voice to a synthesized voice; voice data specifying means for specifying one of voice data stored in the voice data storing means; And sequentially reading the speaking style data in the group searched by the searching means, and using the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speaking style data and the timbre data specified by the timbre data specifying means. Synthesizing means for synthesizing voice, voice selecting means for selecting a desired voice from the voice synthesized by the voice synthesizing means, speech style data and voice color corresponding to the voice selected using the voice selecting means Utterance document storage means for storing data as an utterance document in correspondence with a character string input via the character string input means, thereby providing character information and sound information. The information of the two forms of expression can be created composite information in the rounded shape integrity (spoken documents) that information. In addition, information (spoken document) that matches the character information with the voice information (speech data) including the way of speech intended by the document creator is created.
Expressions such as emotions corresponding to non-verbal information can be added to the document. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

Further, the utterance document creation device of the present invention (claim 6) is characterized in that character string information composed of words, phrases or sentences, phoneme sequence information composed of phonemes corresponding to the characters of the character string information, Speech style composed of duration time of each phoneme, pitch information specifying relative pitch at any time of phoneme sequence information, and velocity information specifying relative volume at any time of phoneme sequence information Based on the character string information of the speaking method data, speaking method data storage means for grouping speaking data having the same character string information and different readings and storing the data for each group; Using a character string input means for inputting a character string constituting a sentence, and a character string input via the character string input means, a group having the same character string information as the character string is used as a data storage method for speaking. Search means for retrieving voice data, voice data storage means for storing a plurality of voice data for adding voice to a synthesized voice, and voice data for specifying one of the voice data stored in the voice data storage means. Voice data specified by the voice data designation means by sequentially reading the speaking style data in the group searched by the specifying means and the searching means, and reading the phonetic sequence information, duration time, pitch information, and velocity information of the read speech style data. Voice synthesis means for synthesizing the voice using the voice synthesis means, a voice selection means for selecting a desired voice from the voice synthesized by the voice synthesis means, and a speech method corresponding to the voice selected using the voice selection means Utterance document storage means for storing data and voice data in correspondence with a character string input via the character string input means and storing the utterance document as an utterance document It is possible to create a character information and the composite information the information of the two expression form of audio information in a rounded shape integrity (spoken documents). It also creates information (spoken documents) that matches the text information with the speech information (speech data) including the way the creator intends to speak, and expresses emotions, etc., which correspond to non-verbal information in the document. Can be added. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

Further, the utterance document creation device of the present invention (Claim 7) is provided with the talking style data creating / registering means for creating the talking style data and registering it in the talking style data storage means.
The document creator can create and register desired speech data, thereby enriching the expression of speech (speech) using the utterance document.

Further, the speech document creating apparatus according to the present invention (claim 8) sets a character string information, a phoneme string information, a duration time, a pitch information and a velocity information, which are the information in the speaking style data, to speak. The creator can create and register the desired speaking data because the creator creates the speaking data and registers it in the speaking data storage means, thereby enriching the expression of speech (speaking) using the spoken document. be able to.

Further, the speech document creation device of the present invention (claim 9) can designate reproduction of the speech document, sequentially read the speaking style data in the speech document, and synthesize the voice. Can be easily confirmed.

Further, the utterance document creation device of the present invention (claim 10) can designate an arbitrary character string unit, a sentence unit, a page unit, or the entire utterance document in the utterance document as a range for reproducing the utterance document. Therefore, reproduction and confirmation of the utterance document can be easily performed.

Further, the utterance document creation device according to the present invention (claim 11) displays the utterance document stored in the utterance document storage means using the character string input means, and selects any character of the displayed utterance document. It is possible to specify a column and change or re-enter the specified character string. Using the changed or re-input character string, search by search means, voice data specification by voice data specification means, voice data by voice synthesis means By performing voice synthesis and voice selection by voice selection means,
Since it is possible to change the way of speaking data and voice data corresponding to the designated character string, it is possible to improve the use range and convenience of the utterance document.

Further, the utterance document creating apparatus according to the present invention (claim 12) includes a plurality of timbre data as a male voice, a female voice, a child voice, an old voice, a sharp voice, a clear voice, and a fat voice. Voice, thin, strong, gentle, and mechanical voices each have sensibly identifiable timbre data, thereby improving the range of use and convenience of spoken documents . Also, more diverse voice synthesis can be performed.

The utterance document creation device according to the present invention (claim 13) has a kana-kanji conversion function, and converts a kanji-kana character converted using a kana-kanji conversion function as a character string input by a character example input means. Since the mixed text can be used, the use range and convenience of the utterance document can be improved. Also, it is possible to perform more free document expression.

Further, in the utterance document creation device according to the present invention (claim 14), the speaking style data includes character string information, phoneme string information,
In addition to the duration time, pitch information, and velocity information, it has type information indicating the classification type of the respective speaking style data. When the classification type is specified, it has the same character string information as the input character string. Speaking data having the same type information as the specified classification type in the group is retrieved from the speaking data storage means, the retrieved speaking data is read, and phoneme sequence information of the read speaking data is read. Since the speech is synthesized using the time length, pitch information and velocity information and the timbre data specified by the timbre data specifying means, it is possible to improve the production efficiency and convenience of the utterance document.

Further, the utterance document creation device of the present invention (Claim 15), as the classification type, converts the speech corresponding to each way of speaking data into a Tokyo type pronunciation type, an Osaka type pronunciation type, a Tokushima type pronunciation type and the like. Since the classification is performed according to the region as described above, by specifying the classification type, it is possible to easily create an utterance document that can perform speech synthesis in a manner of speaking in each region.

The speech document creating apparatus according to the present invention (claim 16) converts the speech corresponding to each type of speech data as a classification type such as an elderly-type pronunciation type, a youth-type pronunciation type, and a high-school-type pronunciation type. As described above, the categorization is made according to the age. By specifying the categorization type, it is possible to easily create an utterance document in which speech synthesis can be performed in each age.

Also, in the utterance document creation device according to the present invention (claim 17), the character string input means has a display unit, and converts the voice data designated for each character string of the utterance document into voice data. Since the font of the character string to be displayed and the decoration method are displayed on the display unit based on the display, it is easy to create and change the utterance document, and to easily grasp the voice data designation status. And the convenience is improved.

The speech document creating method according to the present invention (claim 18) includes a first step of inputting a word, a phrase or a character string constituting a sentence, a character string information comprising a word, a phrase or a sentence, and a character string. Each of phoneme string information composed of phonemes corresponding to the characters of the string information, duration time of each phoneme in the phoneme string information, pitch information specifying the relative pitch of the phoneme string information at an arbitrary time, and phoneme string information Based on the character string information of the speaking data, the speaking data composed of the velocity information specifying the volume of the phoneme is grouped based on the character string information of the speaking data, and the speaking data having the same character string information but different readings are grouped. A second step of searching for a group having the same character string information as the character string input in the first step with reference to the database stored in the first step, and generating voice data for adding voice to the synthesized voice. A third step of constant, sequentially reads out speaking how data in the group searched in the second step,
A fourth step and a fourth step of synthesizing a voice using the read phonetic sequence information, duration time, pitch information and velocity information of the speaking style data, and the timbre data specified in the third step A fifth step of selecting a desired voice from the voice synthesized in the step, and talking data corresponding to the voice selected in the fifth step are stored as a speech document in association with the character string input in the first step. And the sixth step,
It is possible to create information (an utterance document) in which information of two expression forms, that is, character information and voice information, is combined in a consistent manner. It also creates information (spoken documents) that matches the text information with the speech information (speech data) including the way the creator intends to speak, and expresses emotions, etc., which correspond to non-verbal information in the document. Can be added. further,
Speech can be accurately synthesized according to the way of speech intended by the document creator.

The speech document creating method according to the present invention (claim 19) includes a first step of inputting a word, a phrase or a character string constituting a sentence, a character string information comprising a word, a phrase or a sentence, and a character string. Phoneme string information consisting of phonemes corresponding to the characters of the string information, duration of each phoneme in the phoneme string information, pitch information specifying the relative pitch of the phoneme string information at any point in time, and arbitrary phoneme string information Based on the character string information of the speaking data, the speaking data composed of the velocity information specifying the relative volume at the point of time is converted into the speaking data having the same character string information and different readings. With reference to the database that is grouped and stored for each group,
A second step of searching for a group having the same character string information as the character string input in the first step, a third step of specifying voice data for adding a voice to a synthesized voice, and a second step of specifying voice data. The speech style data in the group searched in the process of step 1 is sequentially read, and the phoneme sequence information, duration time, pitch information and velocity information of the read speech style data,
A fourth step of synthesizing a voice using the voice color data specified in the third step, a fifth step of selecting a desired voice from the voice synthesized in the fourth step, and a fifth step of And a sixth step of storing as a spoken document the speech style data corresponding to the voice selected in the process in association with the character string input in the first process, so that the text information and the speech style intended by the document creator are included. (Speech document) can be created with consistency with the voice information (speaking style data) including. In addition, information (spoken document) that matches character information with audio information (speech data) including how the document creator intended to speak.
Can be added to the document to add expressions such as emotions corresponding to non-verbal information. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

Further, according to the speech document creation method of the present invention (claim 20), the reproduction of the speech document is specified in the seventh step of designating the reproduction of the speech document stored in the sixth step and the seventh step. Then, the eighth step of sequentially reading the speech style data and the timbre data in the utterance document and synthesizing the speech is included, so that the utterance document can be easily confirmed.

Further, according to the utterance document creation method of the present invention (claim 21), in the seventh step, the range of reproduction of the utterance document is defined as an arbitrary character string unit, a sentence unit,
Since it is possible to specify a page unit or the entire utterance document, reproduction and confirmation of the utterance document can be easily performed.

The speech document creation method of the present invention (claim 22) displays the speech document stored in the sixth step, specifies an arbitrary character string of the displayed speech document, and A ninth step of performing a change or re-entry.
The second step, the third step, the fourth step, the fifth step, and the sixth step are executed again by using the character string changed or re-input in the step, so that the utterance document is Since the change is possible, the use range and convenience of the utterance document can be improved.

A computer-readable recording medium storing a program for causing a computer to execute the speech document creating procedure according to the present invention is a first medium for inputting a word, a phrase or a character string constituting a sentence. Procedures, character string information consisting of words, phrases or sentences, phoneme string information consisting of phonemes corresponding to the characters of the string information, duration of each phoneme in the phoneme string information, Based on the character string information of the speaking data, speech data composed of pitch information specifying a relative pitch and velocity information specifying the volume of each phoneme in the phoneme sequence information is obtained.
A group having the same character string information as the character string input in the first procedure by referring to a database in which the character string information is the same and the reading data is grouped and the reading is different, and stored in each group. , A third procedure of specifying voice data to add voice to the synthesized voice, and a method of sequentially reading the talking style data in the group searched in the second procedure, and reading the read speech. Using the phonemic sequence information, duration time, pitch information, and velocity information of the voice data and the timbre data specified in the third procedure, the speech is synthesized by the fourth procedure and the fourth procedure. A fifth procedure for selecting a desired voice from the voices obtained from the
Speaking data corresponding to the voice selected in the fifth step is 1
And the sixth step of storing as an utterance document in correspondence with the character string input in the above procedure, so that the consistency between the character information and the speech information (speaking style data) including the way of speech intended by the document creator is included. (Utterance document) can be created. It also creates information (spoken documents) that matches the text information with the speech information (speech data) including the way the creator intends to speak, and expresses emotions, etc., which correspond to non-verbal information in the document. Can be added. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

A computer-readable recording medium storing a program for causing a computer to execute the speech document creating procedure of the present invention is a first medium for inputting a word, a phrase or a character string constituting a sentence. Procedures, character string information consisting of words, phrases or sentences, phoneme string information consisting of phonemes corresponding to the characters of the string information, duration of each phoneme in the phoneme string information, Pitch information specifying relative pitch and phoneme string information Speaking data composed of velocity information specifying relative volume at any point in time is converted into the character string based on the character string information of the talking data. The same sentence as the character string input in the first procedure is referred to by referring to a database in which speaking information data having the same information but different readings is grouped and stored for each group. A second procedure for searching for a group having column information, a third procedure for specifying timbre data for adding a timbre to a synthesized voice, and a method of sequentially reading speaking style data in the group searched for in the second procedure A fourth procedure of synthesizing speech using the phoneme sequence information, duration time, pitch information and velocity information of the read speech style data and the timbre data specified in the third procedure;
5th step of selecting a desired voice from voices synthesized in the procedure of
And the sixth procedure of storing the speaking style data corresponding to the voice selected in the fifth procedure as the utterance document in association with the character string input in the first procedure. It is possible to create information (an utterance document) that is consistent with audio information (speaking style data) including a way of speaking intended by the creator. It also creates information (spoken documents) that matches the text information with the speech information (speech data) including the way the creator intends to speak, and expresses emotions, etc., which correspond to non-verbal information in the document. Can be added. Further, the speech can be accurately synthesized according to the way of speech intended by the document creator.

A computer-readable recording medium storing a program for causing a computer to execute the utterance document creation procedure of the present invention is a computer-readable recording medium storing a utterance document stored in the sixth procedure. When the reproduction of the utterance document is designated in the seventh step, and the eighth step of sequentially reading the speech style data and the timbre data in the utterance document and synthesizing the voice, the utterance document Can be easily confirmed.

A computer-readable recording medium storing a program for causing a computer to execute the utterance document creation procedure according to the present invention is characterized in that, in the seventh procedure, the utterance document creation range is set as the utterance document reproduction range. Since it is possible to specify an arbitrary character string unit, a sentence unit, a page unit, or the entire utterance document, the reproduction and confirmation of the utterance document can be easily performed.

A computer-readable recording medium storing a program for causing a computer to execute the speech document creation procedure of the present invention displays the speech document stored in the sixth procedure, and displays the displayed speech document. A ninth procedure of specifying an arbitrary character string of the document and changing or re-inputting the specified character string, and using the character string changed or re-input in the ninth procedure, By executing the above procedure, the third procedure, the fourth procedure, the fifth procedure, and the sixth procedure, the utterance document can be changed, so that the use range and convenience of the utterance document are improved. be able to.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of an utterance document creation device according to a first embodiment.

FIG. 2 is an explanatory diagram showing talking style data stored in a talking style data storage unit according to the first embodiment;

FIG. 3 is an explanatory diagram showing types of voice data stored in a voice data storage unit according to the first embodiment.

FIG. 4 is an external view of an utterance document creation device according to the first embodiment.

FIG. 5 is a schematic flowchart of an utterance document creation process according to the first embodiment;

FIG. 6 is an explanatory diagram illustrating an example of a display screen of a display unit in an utterance document creation process.

FIG. 7 is an explanatory diagram illustrating an example of a display screen of a display unit in a process of creating an utterance document.

FIG. 8 is an explanatory diagram showing a screen display example of an utterance document created in an utterance document creation process.

FIG. 9 is an explanatory diagram illustrating an example of utterance document data stored in an utterance document storage unit.

FIG. 10 is a schematic flowchart of an utterance document reproduction process according to the first embodiment;

FIG. 11 is an explanatory diagram illustrating an example of a display screen of a display unit in a reproduction process of an utterance document.

FIG. 12 is an explanatory diagram illustrating an example of a display screen of a display unit in a process of reproducing an utterance document.

FIG. 13 is an explanatory diagram illustrating an example of a display screen of a display unit in a reproduction process of an utterance document.

FIG. 14 is a schematic flowchart of an utterance document creation process using type information according to the first embodiment;

FIG. 15 is an explanatory diagram illustrating an example of a display screen of a display unit in a process of creating an utterance document using type information.

FIG. 16 is a schematic flowchart of a reproduction process of an utterance document using type information according to the first embodiment.

FIG. 17 is a diagram illustrating how to create and generate speaking style data according to the first embodiment.
It is a schematic flowchart of a registration process.

FIG. 18 is an explanatory diagram illustrating a display screen of a process of creating and registering speaking style data.

FIG. 19 is an explanatory diagram illustrating a display screen of a process for creating and registering speaking style data.

FIG. 20 is an explanatory diagram showing a display screen of processing for creating and registering speaking style data.

FIG. 21 is an explanatory diagram showing a display screen of processing for creating and registering speaking style data.

FIG. 22 is an explanatory diagram showing a display screen of processing for creating and registering speaking style data.

FIG. 23 is an explanatory diagram showing a display screen of a process of creating and registering speaking style data.

FIG. 24 is an explanatory diagram showing a display screen of processing for creating and registering speaking style data.

FIG. 25 is an explanatory diagram showing a display screen of a process of creating and registering speaking style data.

FIG. 26 is a schematic flowchart of an utterance document change process according to the first embodiment;

FIG. 27 is a schematic flowchart of an utterance document creation process according to the second embodiment.

FIG. 28 is a schematic flowchart of a process of changing information in speaking style data according to the second embodiment.

FIG. 29 is an explanatory diagram showing a display screen of a process of changing information in speaking style data according to the second embodiment.

FIG. 30 is an explanatory diagram showing a display screen of a process of changing information in the talking style data according to the second embodiment.

[Explanation of symbols]

 101 control unit 101a CPU 101b ROM 101c RAM 102 application storage unit 103 speech data storage unit 104 voice data storage unit 105 voice synthesis unit 106 key input unit 107 display unit 108 microphone 109 speaker 110 utterance document storage unit 111 interface (I / F) 112 FD drive 113 CD-ROM drive 114 Communication unit 201 Speaking data 202 Character string information 203 Phoneme sequence information 204 Duration length 205 Pitch information 206 Velocity information 207 Type information

Continuation of front page (56) References JP-A-62-217294 (JP, A) JP-A-2-234198 (JP, A) JP-A-2-247696 (JP, A) JP-A-3-89300 (JP) JP-A-4-20998 (JP, A) JP-A-7-134597 (JP, A) JP-A-7-319495 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB G10L 13/00-13/08 G06F 3 / 16,17 / 21

Claims (27)

(57) [Claims] 1. Character string information composed of a word, a phrase or a sentence, phoneme string information composed of phonemes corresponding to characters of the character string information, duration time of each phoneme in the phoneme string information, Based on the character string information of the speaking data, speech data composed of pitch information designating a relative pitch at an arbitrary point in time and velocity information designating a volume of each phoneme of the phoneme sequence information is obtained. Speaking data storage means for grouping speaking data having the same character string information and different readings and storing the data for each group, and character string input for inputting a character string constituting a word, a phrase or a sentence Means for searching a group having the same character string information as the character string from the speaking data storage means using the character string input via the character string input means; Voice data storage means for storing a plurality of voice data for adding voice to the voice to be synthesized; and sequentially reading the speaking data in the group searched by the searching means, phonemic sequence information of the read talking data, Voice synthesis means for synthesizing voice using the duration time, pitch information and velocity information and one of the voice data stored in the voice data storage means; and a desired voice from the voice synthesized by the voice synthesis means. A voice selecting means for selecting a voice, and speaking style data corresponding to the voice selected using the voice selecting means are stored as an utterance document in association with a character string input via the character string input means. An utterance document creation device, comprising: utterance document storage means. 2. A phoneme string information comprising a word, a phrase or a sentence, a phoneme string information comprising a phoneme corresponding to a character of the character string information, a duration time of each phoneme in the phoneme string information, A speech string composed of pitch information specifying a relative pitch at an arbitrary point in time and velocity information specifying a relative volume of the phoneme string information at an arbitrary point in time, Based on the information, the speaking style data storage means for grouping speaking style data having the same character string information and different readings and storing the data for each group, and a character string constituting a word, a phrase or a sentence are input. Character string input means for searching for a group having the same character string information as the character string from the speaking data storage means using the character string input through the character string input means. Searching means; voice data storing means for storing a plurality of voice data for adding voice to the voice to be synthesized; and sequentially reading the talking data in the group searched by the searching means. Voice synthesis means for synthesizing voice using phoneme string information, duration time, pitch information and velocity information and one of the voice data stored in the voice data storage means, and voice synthesized by the voice synthesis means A voice selecting means for selecting a desired voice from the user; and speaking the speech data corresponding to the voice selected using the voice selecting means in correspondence with the character string input through the character string input means. An utterance document creation device, comprising: utterance document storage means for storing as a document. 3. A reproduction designating means for designating reproduction of an utterance document stored in the utterance document storage means, and when reproduction of the utterance document is designated, the speech synthesis means causes the utterance document to be reproduced. 3. The method according to claim 1, wherein the speech data is read out sequentially to synthesize speech.
The described utterance document creation device. 4. The reproduction specifying means is capable of specifying an arbitrary character string unit, a text unit, a page unit, or the entire utterance document in the utterance document as a range in which the utterance document is reproduced. The utterance document creation device according to claim 3. 5. Character string information composed of a word, a phrase or a sentence, phoneme string information composed of phonemes corresponding to characters of the character string information, duration time of each phoneme in the phoneme string information, Based on the character string information of the speaking data, speech data composed of pitch information designating a relative pitch at an arbitrary point in time and velocity information designating a volume of each phoneme of the phoneme sequence information is obtained. Speaking data storage means for grouping speaking data having the same character string information and different readings and storing the data for each group, and character string input for inputting a character string constituting a word, a phrase or a sentence Means for searching a group having the same character string information as the character string from the speaking data storage means using the character string input via the character string input means; Voice data storing means for storing a plurality of voice data for adding voice to a voice to be synthesized; voice data specifying means for specifying one of voice data stored in the voice data storing means; Means for sequentially reading the speech data in the group searched by the means, and using the phoneme sequence information, duration time, pitch information and velocity information of the read speech data and the voice data specified by the voice data specifying means. Voice synthesis means for synthesizing voice by using the voice synthesis means, voice selection means for selecting a desired voice from the voice synthesized by the voice synthesis means, and speaking style data corresponding to the voice selected using the voice selection means And an utterance document storage means for storing voice data as an utterance document in association with a character string input via the character string input means. Spoken document creating apparatus characterized by. 6. Character string information composed of a word, a phrase or a sentence, phoneme string information composed of phonemes corresponding to the characters of the character string information, duration time of each phoneme in the phoneme string information, A speech string composed of pitch information specifying a relative pitch at an arbitrary point in time and velocity information specifying a relative volume of the phoneme string information at an arbitrary point in time, Based on the information, the speaking style data storage means for grouping speaking style data having the same character string information and different readings and storing the data for each group, and a character string constituting a word, a phrase or a sentence are input. Character string input means for searching for a group having the same character string information as the character string from the speaking data storage means using the character string input through the character string input means. Searching means; voice data storing means for storing a plurality of voice data for adding voice to voice to be synthesized; voice data specifying means for specifying one of voice data stored in the voice data storing means And sequentially reading out the speaking style data in the group searched by the searching means, the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speaking style data, and the voice specified by the voice color data specifying means. Voice synthesizing means for synthesizing voice using data, voice selecting means for selecting a desired voice from voice synthesized by the voice synthesizing means, and corresponding to voice selected using the voice selecting means. An utterance document storage device that stores speech data and voice data to be spoken in association with a character string input via the character string input means as an utterance document. An utterance document creation device, comprising: a step; 7. A method for creating speaking data, which creates the speaking data and registers the data in the speaking data storage means.
7. The utterance document creation device according to claim 5, further comprising registration means. 8. The speech style data creation / registration means, which inputs speech waveform data recorded in advance or a real voice uttered by a user and displays speech waveform data, and speech data input means for displaying speech waveform data; Phoneme analysis of the speech waveform data, setting and displaying the duration of each phoneme, and phoneme sequence information adding means for adding phoneme sequence information corresponding to the set duration. Pitch curve display means for analyzing the pitch of the speech waveform data to display a pitch curve, and a relative pitch of the phoneme string information at an arbitrary point in time based on the displayed pitch curve and phoneme string information. Pitch information generating means for generating pitch information by adjusting or adding the value of pitch information; and velocity information for generating velocity information by adjusting the volume of each phoneme in the phoneme sequence information. Character information generating means, character string information setting means for setting character string information by inputting a character string corresponding to the voice waveform data, and a corresponding group in the speaking style data storage means based on the character string information. 8. The utterance document creating apparatus according to claim 7, further comprising a registration unit for registering the character string information, phoneme string information, duration time, pitch information, and velocity information as speaking style data. 9. A method according to claim 1, further comprising a reproduction designating means for designating reproduction of the utterance document stored in said utterance document storage means. 9. The utterance document creation apparatus according to claim 5, wherein the speech data and the timbre data in the speech data are sequentially read out to synthesize a speech. 10. The reproduction specifying means is capable of specifying an arbitrary character string unit, a sentence unit, a page unit, or the entire utterance document in the utterance document as a range in which the utterance document is reproduced. The utterance document creation device according to claim 9. 11. An utterance document stored in the utterance document storage means is displayed by using the character string input means, and an arbitrary character string of the displayed utterance document is designated. A column can be changed or re-input. Using the changed or re-input character string, a search by the search means, a specification of voice data by the voice data specification means, a synthesis of voice by the voice synthesis means, By selecting a voice by the voice selecting means,
The speech style data and the timbre data corresponding to the specified character string can be changed.
10. The utterance document creation device according to claim 10. 12. The plurality of timbre data is a male voice,
Women's voice, children's voice, old man's voice, sharpened voice, clear voice,
7. The utterance document creation according to claim 1, wherein the utterance data is sensibly identifiable voice data such as a thick voice, a thin voice, a powerful voice, a gentle voice, and a mechanical voice. apparatus. 13. The character string input means has a kana-kanji conversion function, and the character string input by the character example input means is
4. A text with kanji and kana mixed after being converted using the kana-kanji conversion function.
An utterance document creation device according to 5 or 6. 14. A classification type designating means for designating a classification type of the speaking style data, wherein the speaking style data is added to the character string information, phoneme string information, duration time, pitch information and velocity information. hand,
When the classification type is specified via the classification type specifying means, the search means includes a character string input via the character string input means. And using the classification type specified through the classification type specifying means, a group having the same character string information as the character string, and speaking style data having the same type information as the specified classification type From the speaking data storage means, and the speech synthesizing means reads the speaking data retrieved by the searching means, and obtains phoneme string information, duration time, pitch information and velocity information of the read speaking data. 7. An utterance document creating apparatus according to claim 5, wherein a voice is synthesized using the voice data specified by the voice data specifying means. . 15. The classification type is characterized in that voices corresponding to respective manners of speech data are classified according to region, such as Tokyo-type pronunciation type, Osaka-type pronunciation type, and Tokushima-type pronunciation type. The utterance document creation device according to claim 14, wherein: 16. The classification type is characterized in that voices corresponding to respective manners of speech data are classified according to age, such as an elderly-type pronunciation type, a youth-type pronunciation type, and a high-school-type pronunciation type. The utterance document creation device according to claim 14, wherein: 17. The character string input means has a display unit and, based on voice data specified for each character string of the utterance document, a font of a character string to be displayed or a decoration. 17. The utterance document creating apparatus according to claim 5, wherein the utterance document is displayed on the display unit by changing a method. 18. A first step of inputting a character string constituting a word, a phrase or a sentence, character string information comprising a word, a phrase or a sentence, and phoneme sequence information comprising a phoneme corresponding to a character of the character string information. , The duration of each phoneme in the phoneme string information, pitch information specifying the relative pitch of the phoneme string information at any point in time, and velocity information specifying the volume of each phoneme in the phoneme string information. Based on the character string information of the speaking data, the speaking data to be spoken are referred to a database in which the speaking data having the same character string information and different readings are grouped and stored for each group, A second step of searching for a group having the same character string information as the character string input in the first step, and a third step of specifying voice data to add voice to the synthesized voice. The speech style data in the group searched in the second step is sequentially read, and the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speech style data and the speech information specified in the third step are read. A fourth step of synthesizing the voice using the timbre data; a fifth step of selecting a desired voice from the voice synthesized in the fourth step; and a step of synthesizing the voice selected in the fifth step. And a sixth step of storing the corresponding way of speaking data as the utterance document in association with the character string input in the first step. 19. A first step of inputting a character string constituting a word, a phrase or a sentence, character string information comprising a word, a phrase or a sentence, and phoneme sequence information comprising a phoneme corresponding to a character of the character string information. , The duration of each phoneme in the phoneme string information, the pitch information specifying the relative pitch of the phoneme string information at any time, and the relative volume of the phoneme string information at any time. A database in which speaking style data composed of velocity information to be read is grouped based on character string information of the speaking style data, and speaking style data having the same character string information and different readings are grouped and stored for each group. , A second step of searching for a group having the same character string information as the character string input in the first step, and specifying voice data to add a voice to the synthesized voice. A third step, sequentially reading the speaking style data in the group searched in the second step, and reading phonemic sequence information, duration time, pitch information and velocity information of the read speaking style data; A fourth step of synthesizing a voice using the voice color data specified in the step, a fifth step of selecting a desired voice from the voice synthesized in the fourth step, and a fifth step of And a sixth step of storing the way of speaking data corresponding to the voice selected in the step as the utterance document in association with the character string input in the first step. 20. A seventh step of designating the reproduction of the utterance document stored in the sixth step, and the reproduction of the utterance document is designated in the seventh step.
20. An utterance document creation method according to claim 18 or 19, further comprising: an eighth step of sequentially reading speech style data and voice color data in the utterance document and synthesizing a voice. 21. In the seventh step, an arbitrary character string unit, a sentence unit, a page unit, or the entire utterance document in the utterance document can be specified as a range in which the utterance document is reproduced. 21. The method according to claim 20, wherein: 22. A ninth step of displaying the utterance document stored in the sixth step, specifying an arbitrary character string of the displayed utterance document, and changing or re-inputting the specified character string. Using the character string changed or re-input in the ninth step, again using the second step, the third step, the fourth step,
22. The method according to any one of claims 18 to 21, wherein the utterance document can be changed by executing the fifth step and the sixth step. 23. A first procedure for inputting a character string constituting a word, a phrase or a sentence, character string information comprising a word, a phrase or a sentence, and phoneme sequence information comprising a phoneme corresponding to a character of the character string information. , The duration of each phoneme in the phoneme string information, pitch information specifying the relative pitch of the phoneme string information at any point in time, and velocity information specifying the volume of each phoneme in the phoneme string information. Based on the character string information of the speaking data, the speaking data to be spoken are referred to a database in which the speaking data having the same character string information and different readings are grouped and stored for each group, A second procedure for searching for a group having the same character string information as the character string input in the first procedure, a third procedure for specifying voice data for adding voice to the synthesized voice, The speech style data in the group searched in the second procedure is sequentially read out, and the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speech style data and the speech information designated in the third procedure are read out. A fourth procedure of synthesizing a voice using the voice color data, a fifth procedure of selecting a desired voice from the voice synthesized in the fourth procedure, and a voice selected in the fifth procedure. A computer-readable recording medium storing a program for causing a computer to execute a sixth procedure of storing the corresponding manner of speaking data as an utterance document in association with the character string input in the first procedure; . 24. A first procedure for inputting a character string constituting a word, a phrase or a sentence, character string information comprising a word, a phrase or a sentence, and phoneme string information comprising a phoneme corresponding to a character of the character string information. , The duration of each phoneme in the phoneme string information, the pitch information specifying the relative pitch of the phoneme string information at any time, and the relative volume at any time of the phoneme string information. A database in which speaking style data composed of velocity information is grouped based on character string information of the speaking style data and speaking style data having the same character string information but different readings is grouped and stored for each group. In reference to,
A second procedure of searching for a group having the same character string information as the character string input in the first procedure, a third procedure of specifying voice data for adding voice to a synthesized voice, The speech style data in the group searched in the second procedure is sequentially read, and the phoneme sequence information, the duration time, the pitch information and the velocity information of the read speech style data, and the voice specified in the third procedure are read out. A fourth procedure for synthesizing the voice using the data, a fifth procedure for selecting a desired voice from the voice synthesized in the fourth procedure, and a voice corresponding to the voice selected in the fifth procedure. A sixth step of storing as a speech document the speech style data to be associated with the character string input in the first step, and a program for causing a computer to execute Capacity recording medium. 25. A seventh procedure for designating the reproduction of the utterance document stored in the sixth procedure, and when the reproduction of the utterance document is designated in the seventh procedure,
25. A computer readable program according to claim 23 or 24, wherein a program for causing a computer to execute an eighth procedure of sequentially reading out the speaking style data and the timbre data in the utterance document and synthesizing a voice is stored. Recording medium. 26. The method according to claim 7, wherein in the seventh procedure, an arbitrary character string unit, a sentence unit, a page unit, or the entire utterance document in the utterance document can be designated as a reproduction range of the utterance document. 26. The computer-readable recording medium according to claim 25, wherein: 27. A ninth step of displaying the utterance document stored in the sixth step, specifying an arbitrary character string of the displayed utterance document, and changing or re-inputting the specified character string. Using the character string changed or re-input in the ninth procedure, again using the second procedure, the third procedure, the fourth procedure,
27. The computer-readable recording medium according to claim 23, wherein the utterance document can be changed by executing a fifth procedure and a sixth procedure.